[Bee] Docling supports embedded image detection for docx and pptx to deal with hybrid documents

### Requested feature
Docling has a simplepipeline to do PPTX/docx file processing, but it can only extract programmatic text, which means it does not support hybrid processing. So:
**Feature Request:**
1. Add embedded image detection function
2. Support hybrid processing, if ocr is enabled, convert pptx/docx document to PDF with `LibreOffice` and do the text extraction. (If this is too much, request 1 should be added)
...

**Expected Behavior:**
Two processing paths should be there based on the embedded images in the file, however, `docling does not have a function to detect the images before the text extraction`

...

related comments in an open ticket: https://github.com/docling-project/docling/issues/2225#issuecomment-3264970983